®®®® SIIA Público

Título del libro: Computer Methods For Geomechanics: Frontiers And New Applications
Título del capítulo: An attenuation tree for PGA and spectral ordinates

Autores UNAM:
SILVIA RAQUEL GARCIA BENITEZ; MIGUEL PEDRO ROMO ORGANISTA;
Autores externos:

Idioma:
Inglés
Año de publicación:
2011
Palabras clave:

Advanced learning; Geotechnical conditions; Limited observations; Machine learning methods; Partial knowledge; Response spectra; Seismic phenomena; Spectral ordinates; Forestry; Geomechanics; Learning systems; Trees (mathematics); Forestry; Mathematics; Trees


Resumen:

This article investigates a machine learning method, regression trees, to enhance current groundmotion relations. A regression tree represents past experiences and tries to discover patterns from which reasonable predictions regarding future conditions can be made. The attenuation tree is constructed using an extensive world database and the fi nal scheme, extremely simple, predicts accurately the damaging levels of motion for future earthquakes (in terms of PGA and response spectra). This advanced learning tool can be used to describe the dependencies between the input quantities that govern the physical process. The artifi cial intelligence scheme seems to be a very promising alternative to describe the seismic phenomena despite of the limited observations, judgments, partial knowledge of geotechnical conditions, and ambiguous reasoning employed to infer the behavior of the ground motion phenomena universe.


Entidades citadas de la UNAM: